OSPREY

OSPREY performs structure-based protein design by combining dead-end elimination (DEE) and A* tree search to identify global minimum energy conformations (GMEC) under a rigid-backbone, discrete side-chain rotamer model.

Key Features:

• DEE and A* search: Implements dead-end elimination (DEE) together with A* tree search algorithms to explore sequence and rotamer space.
• Rigid-backbone discrete-rotamer model: Assumes rigid backbones with discrete side-chain conformations to enable efficient conformational enumeration.
• GMEC identification: Uses A* search with heuristic guidance to identify the global minimum energy conformation (GMEC).
• Heuristic function optimization: Optimizes heuristic functions within A* to improve the efficiency of GMEC searches.
• GPU parallelization: Implements a parallel A* variant for massively parallel processing on a single GPU, accelerating the design process by up to four orders of magnitude compared to traditional methods.
• Memory management: Incorporates memory-management strategies to address A* search memory constraints while preserving computational speed.
• Integration with iMinDEE and continuous side-chain flexibility: Integrates iMinDEE for rotamer pruning and to accommodate continuous side-chain flexibility during design.

Scientific Applications:

• Protein Stability Enhancement: Designing proteins that maintain their structure and function under diverse conditions.
• Substrate Specificity Alteration: Engineering enzymes or binding proteins to change substrate selectivity.

Methodology:

Uses dead-end elimination (DEE) and A* tree search under a rigid-backbone/discrete-rotamer model, with heuristic function optimization, a GPU-parallel A* variant, memory-management techniques for A* search, and integration with iMinDEE for rotamer pruning and continuous side-chain flexibility to identify the GMEC.